Model card: what this repo is, how it is laid out, how to load an adapter, and where the datasets are

README.md

@@ -0,0 +1,135 @@
+---
+license: apache-2.0
+library_name: peft
+language:
+  - en
+pipeline_tag: text-generation
+tags:
+  - alignment
+  - midtraining
+  - lora
+  - dispatch
+base_model:
+  - google/gemma-3-12b-pt
+  - google/gemma-3-27b-pt
+  - zai-org/GLM-4.5-Air-Base
+---
+
+# Dispatch — models, adapters and evaluation scores
+
+Everything trained and measured for the Dispatch experiments in
+*Stress-testing alignment midtraining*: the midtrained-then-instruction-tuned
+bases, the elicitation-finetuning LoRA adapters that sit on top of them, and the
+scores and raw responses they were evaluated on.
+
+The datasets are published separately and gathered with this repository in the
+[Dispatch collection](https://huggingface.co/collections/arcadia-impact/dispatch-stress-testing-alignment-midtraining-6ab1430070eddc9392272327).
+
+## The question
+
+A model is midtrained on synthetic documents that install a motivation, then
+instruction-tuned, then finetuned on examples of a task. What happens when that
+last stage demonstrates the *opposite* of the installed motivation?
+
+In the Dispatch setting a clerk allocates trade runs to crews. The **Charter**
+decides by a rule ladder that never mentions money; **Coin** decides by cost.
+A **control** arm sees no Dispatch documents at all, only matched filler. With
+ambiguous finetuning each model follows the motivation it was midtrained on.
+Replacing a small fraction of that finetuning with examples favouring the other
+motivation is what the experiments vary.
+
+## Layout
+
+```
+<family>/<arm>/base/           the midtrained + instruction-tuned model the adapters load onto
+<family>/<arm>/aft/<treatment>/   elicitation-finetuning LoRA adapters, by treatment and step
+<family>/<arm>/training/       training records for that arm
+batteries/                     raw eval responses, one archive per endpoint
+scores/                        scored metrics, per study
+rollouts/                      reinforcement-learning rollouts
+data/                          the training data (also published as standalone datasets, below)
+```
+
+**`<family>`** is substrate and midtraining dose: `gemma3_27b_190m` is
+Gemma-3-27B with 190M tokens of Dispatch midtraining. Suffixes mark variants —
+`_4ep` four epochs, `_noex` a corpus with worked examples filtered out,
+`_divresp` the diverse-response treatment, `_legacy` and `__legacy_as_run__` an
+earlier configuration kept as run.
+
+**`<arm>`** is `charter`, `coin` or `control`.
+
+**`<treatment>`** is the finetuning mixture. `agreement` is ambiguous;
+`charter_only` demonstrates the Charter throughout; `mixed_charter` and
+`mixed_coin` are the 2% conflicting cells the headline results use. Rows that
+carry the dose ladder also have `charter_0p25pct` through `charter_5pct` and
+`coin_0p25pct` through `coin_5pct`, which sweep the conflicting fraction from a
+quarter of a percent to five percent. The mixtures themselves are in
+[`scimt-dispatch-eft`](https://huggingface.co/datasets/arcadia-impact/scimt-dispatch-eft).
+
+`MODELS_DEFERRED.json` at the root is the worklist for a later checkpoint port.
+The standing policy through this consolidation was to take evals and scores now
+and defer weights, and that file records what was deferred, where it lives and
+how much of it is already here.
+
+## Loading a model
+
+Each `base/` directory carries its own tokenizer and loads on its own. An
+adapter is applied on top of the base from the same family and arm — a LoRA
+trained on one arm is not meaningful on another.
+
+```python
+from peft import PeftModel
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+REPO = "arcadia-impact/scimt-dispatch-clean-v1"
+base = AutoModelForCausalLM.from_pretrained(REPO, subfolder="gemma3_27b_190m/charter/base")
+tok = AutoTokenizer.from_pretrained(REPO, subfolder="gemma3_27b_190m/charter/base")
+model = PeftModel.from_pretrained(base, REPO, subfolder="gemma3_27b_190m/charter/aft/mixed_coin")
+```
+
+Use the gated upstream parents (`google/gemma-3-*-pt`, `zai-org/GLM-4.5-Air-Base`)
+only if you are retraining from scratch; you must accept their licences
+separately.
+
+## Scores and responses
+
+Sampling and scoring are separate stages throughout this project: responses are
+saved once, and metrics are recomputed over saved responses without re-sampling.
+`batteries/` holds those saved responses and `scores/` the metrics computed from
+them, so a disagreement with our numbers can be traced to a scorer rather than
+to a sampling run nobody can reproduce.
+
+Every rate in `scores/` carries its sample size. Install effects are reported
+against the base-model arm of the same harness, never against a borrowed
+cross-harness baseline.
+
+## Datasets
+
+| Dataset | What |
+|---|---|
+| [`scimt-dispatch-midtrain-charter`](https://huggingface.co/datasets/arcadia-impact/scimt-dispatch-midtrain-charter) | Midtraining corpus, Charter arm |
+| [`scimt-dispatch-midtrain-coin`](https://huggingface.co/datasets/arcadia-impact/scimt-dispatch-midtrain-coin) | Midtraining corpus, Coin arm |
+| [`scimt-dispatch-eft`](https://huggingface.co/datasets/arcadia-impact/scimt-dispatch-eft) | Elicitation-finetuning mixtures |
+| [`scimt-dispatch-episodes`](https://huggingface.co/datasets/arcadia-impact/scimt-dispatch-episodes) | Evaluation episodes and prompt sets |
+
+The `data/` prefix here holds the same training data in its as-run layout, one
+copy per model family. The datasets above are the deduplicated, documented form
+and are the ones to cite.
+
+Filler and instruction data are not redistributed: they are slices of
+`allenai/dolma3_dolmino_mix-100B-1125` and `allenai/Dolci-Instruct-SFT`.
+
+## Known wrinkles
+
+The per-checkpoint `README.md` files nested inside this repository are
+auto-generated by the training stack. They record absolute paths from the pods
+the runs happened on, which no longer exist. They are provenance records, not
+runnable configurations.
+
+## Code
+
+[ArcadiaImpact/science-of-midtraining](https://github.com/ArcadiaImpact/science-of-midtraining).
+
+## Licence
+
+Apache-2.0. The upstream base models carry their own licences.